Reliability of LLMs as medical assistants for the general public: a randomized preregistered study

February 1, 2026
AI
Bean, Andrew M., Payne, Rebecca Elizabeth, Parsons, Guy, Kirk, Hannah Rose, Ciro, Juan, Mosquera-Gómez, Rafael, Hincapié M, Sara, Ekanayaka, Aruna S., Tarassenko, Lionel, Rocher, Luc, Mahdi, Adam
Pathogen:N/A
Infection Type:N/A
Pathogen Type:N/A

Summary

This article describes a randomized preregistered study evaluating the reliability of Large Language Models (LLMs) when used as medical assistants by the general public. The study aims to assess how accurately LLMs provide medical information and advice to lay users, focusing on their potential utility and limitations in a public health context. The research likely investigates various medical queries and LLM responses to determine their safety and trustworthiness for non-experts seeking health guidance. This is a study on the reliability of AI, not a clinical case report about a patient's infection.

Key note: This article is a study on the reliability of LLMs in medicine, not a clinical case report about an infection.

DOI: 10.1038/s41591-025-04074-y